In the recent years machine learning techniques have attracted the attention of wind energy community to make use of the large amount of available data produced from the running wind turbines. These modern wind turbines are typically equipped with measurement systems and sensors that can provide a wealth of information about the operating conditions of the machine. Nevertheless, not all the acquired raw data can be used effectively to enhance the operation of a turbine. This work addresses the question of estimating the damage equivalent loads (DEL) of different components of a drivetrain. The estimation is based on low frequency sampled typically available SCADA measurements. Typical SCADA measurements that are used as input for the estimation model are generator rotational speed, low speed shaft torque and generator torque as well as, wind speed and direction. Several machine learning methods as random forests (RF), support vector machines (SVM), linear regression (LR), decision trees and neural networks (NN) were developed, exhibiting different behavior for each approach. The qualitative and quantitative performance of each algorithm are evaluated and compared against each other. Furthermore, analysis of importance of the input features is presented.

The inevitable transition of the power system toward a sustainable and renewable-energy centered power system is accompanied by huge versatility and significant challenges. A corresponding shift in operation strategies, embracing more intelligence and digitization, e.g., a Cyber-Physical System (CPS), is needed to achieve an optimal, reliable and secure operation across all system levels (components, units, plants, grids) and by the use of big data. Digital twins (DTs) are a promising approach to realize CPS. In this paper, their applications in power systems are reviewed comprehensively. The review reveals that there exists a gap between available DT definitions and the requirements for DTs utilized in future power systems. Therefore, by adapting the current definitions to these requirements, a generic definition of a “Digital Twin System (DTS)” is introduced which finally allows proposing a multi-level and arbitrarily extendable “System of Digital Twin Systems (SDTSs)” idea. The SDTSs can be realized with an open-source framework that serves as a central data and communication interface between different DTSs which can interact by “Reporting Modules” and are regulated by “Control Modules” (CMs). Exemplary application scenarios involving multiple system levels are discussed to illustrate the capabilities of the proposed SDTS concept.

Data-driven approaches have gained interest recently in the field of wind energy. Data-driven online estimators have been investigated and demonstrated in several applications such as online loads estimation, wake center position estimations, online damage estimation. The present work demonstrates the application of machine learning algorithms to formulate an estimator of the internal loads acting on the bearings of the drivetrain of onshore wind turbines. The loads estimator is implemented as a linear state-space model that is augmented with a non-linear feed-forward neural network. The estimator infers the loads time series as a function of the standard measurements from the SCADA and condition monitoring systems (CMS). A formal analysis of the available data is carried out to define the structure of the virtual sensor regarding the order of the models, number of states, architecture of neural networks. Correlation coefficient of 98% in the time domain and matching of the frequency signature are achieved. Several applications are mentioned and discussed in this work such as online estimation of the forces for monitoring and model predictive control applications.